Fast prots: Investor sentiment and stock returns during Ramadan

Je drzej Biakowski
a,
, Ahmad Etebari
b
, Tomasz Piotr Wisniewski
c
a
University of Canterbury, Department of Economics and Finance, College of Business and Economics, Christchurch, New Zealand
b
University of New Hampshire, Department of Accounting and Finance, Whittemore School of Business and Economics, Durham, NH, USA
c
University of Leicester, School of Management, Leicester, England, United Kingdom
a r t i c l e i n f o
Article history:
Received 25 November 2010
Accepted 19 September 2011
Available online 24 September 2011
JEL classication:
G12
G14
Keywords:
Ramadan effect
Behavioral nance
Market efciency
Religion
a b s t r a c t
Observed by more than 1.5 billion Muslims, Ramadan is one of the most celebrated religious traditions in
the world. We investigate stock returns during Ramadan for 14 predominantly Muslim countries over the
years 19892007. The results show that stock returns during Ramadan are signicantly higher and less
volatile than during the rest of the year. No discernible declines in market liquidity are recorded. We nd
these results consistent with a notion that Ramadan positively affects investor psychology, as it promotes
feelings of solidarity and social identity among Muslims world-wide, leading to optimistic beliefs that
extend to investment decisions.
2011 Elsevier B.V. All rights reserved.
1. Introduction
A growing body of recent research shows that security returns
respond to variables related to factors such as the weather, bio-
rhythms, beliefs, and social identity.
1
For instance, Hirshleifer and
Shumway (2003) nd that daily stock returns are signicantly corre-
lated with sunshine. Dowling and Lucey (2005) and Kamstra et al.
(2000) present international evidence that seasonal variations in
biorhythms and disruptions in sleep caused by changing to and from
daylight saving time affect stock returns. Finally, Edmans et al.
(2007) investigate the impact of international soccer results and nd
a signicant market declines after losses by national soccer teams in
international competitions.
These studies are motivated by a set of recent theories that fo-
cus on the effects of emotions and feelings on peoples judgments
and decision making.
2
For example, Loewenstein et al. (2001) risk-
as-feelings theory posits that peoples feelings and emotions often
inuence their decisions, especially when such decisions involve risk
and uncertainty. According to this theory both emotional reactions
and cognitive evaluations guide reasoning, but when they diverge,
emotional reactions often dominate behavior and inuence the
eventual decision (Simon, 1967; Loewenstein et al., 2001). In line
with the ndings that people in good moods tend to be more opti-
mistic in their judgments than those in bad moods (Wright and
Bower, 1992), these studies demonstrate that market prices can be
inuenced by changes in investors emotional state even when the
underlying events are economically neutral from a direct cost-bene-
t perspective.
In this paper we examine whether a religious practice can,
through its inuence on investors psychology, affect the behav-
ior of the market. The important role played by religion has been
highlighted in several earlier studies. Weber (1905) argue that
Protestantism fueled the development of early capitalism. Stulz
and Williamson (2003) document empirically that religion has
the power to explain the cross-country variation in creditor
rights and the level of enforcement. The existing literature also
acknowledges that religiosity, social interactions and social
norms can inuence investment decisions of individuals and
nancial institutions, as well as corporate decision-making in
general (Duo and Saez, 2002; Hong et al., 2004; Brown and
Taylor, 2010; Hong and Kacperczyk, 2009; Hilary and Hui,
2009). We endeavor to add to the existing body of knowledge
by focusing on the stock markets in countries where religion is
an integral part of everyday life and determines much of the
0378-4266/$ - see front matter 2011 Elsevier B.V. All rights reserved.
doi:10.1016/j.jbankn.2011.09.014

Corresponding author. Tel.: +64 3 364 3316; fax: +64 3 364 2635.
E-mail addresses: jedrzej.bialkowski@canterbury.ac.nz (J. Biakowski), ahmad.
etebari@unh.edu (A. Etebari), tpw5@leicester.ac.uk (T.P. Wisniewski).
1
For a comprehensive review of the theory and evidence on this line of research,
see Hirshleifer (2001), and Lucey and Dowling (2005).
2
The impact of investors mood on their actions is of great interest to researchers in
Behavioral Finance. The survey by Subrahmanyam (2007) reviews the developments
in this eld over the past two decades.
Journal of Banking & Finance 36 (2012) 835845
Contents lists available at SciVerse ScienceDirect
Journal of Banking & Finance
j our nal homepage: www. el sevi er . com/ l ocat e/ j bf
interaction within the society. Specically, we examine the stock
market effects of Ramadan for nations where the majority of the
population are adherents of Islam.
Observed by more than 1.5 billion Muslims, Ramadan is one of
the most celebrated religious traditions in the world. It involves
abstinence from eating, drinking and other sensual pleasures, as
well as strict control of desires. In principle, Ramadan is a time
of reection, self-reformation, giving and worship (Quran 51:21;
2:183). As a fundamental, shared experience, it brings about a
greater solidarity and cooperation among Muslims throughout
the world. A major contributor to this effect is the greater social
support provided within the community and also the closer rela-
tionship an individual Muslim establishes with Allah, which in it-
self is a critical form of social relation. During the holy month
Muslims become more socially and spiritually oriented. As indi-
cated by research in positive psychology, religion provides a valu-
able form of social support, encourages optimistic beliefs, and
contributes to the believers happiness (Beit-Hallahmi and Argyle,
1997). The collective enthusiasm derived from Ramadan leads to
a heightened sense of social identity and greater satisfaction with
life for Muslims around the world. This satisfaction, in turn, may
conceivably lead to optimistic beliefs that extend to the invest-
ment decisions of the followers.
The process of Ramadan-type fasting itself can substantially
benet the health of the devotee. As mentioned in the Quran
and also supported by clinical research, it can promote both the
physical and mental well-being of individuals (Bck et al., 1978;
Perk et al., 2001; Saleh et al., 2005). Clinical research shows that
the Ramadan fasting generally makes people less tense and anx-
ious (Daradkeh, 1992) and that it may also induce mild states of
euphoria (Knerr and Pearl, 2008). The euphoria derived from Ram-
adan could inuence investor behavior and consequently have a
positive valuation effect on equity markets in Islamic countries.
Mood aside, the demand for equities may also increase as a result
of the favorable health status enjoyed by the local Muslim inves-
tors. Rosen and Wu (2004) report that households in good health
are inclined to hold a greater share of risky assets in their portfo-
lios. Similarly, Edwards (2008) nds that retired people perceive
signicant risks associated with future health shocks and that
these perceptions lead to a continued reduction in nancial risk
taking. In light of the foregoing discussion, we would expect the
Ramadan period to coincide with notable increases in the prices
of risky securities.
This study is a comprehensive examination of stock returns for
a broad sample of 129 Ramadan months for 14 predominantly
Muslim countries over the period running from 1989 to 2007.
The results showthat during Ramadan stock returns are on average
much higher but less volatile compared to the rest of the year. The
results also attest to the fact that there are no discernible declines
in market liquidity during Ramadan. We nd these results consis-
tent with our prior expectation that Ramadan has a positive impact
on the mood and hence on investors sentiment. Interestingly,
stock markets located in countries populated by non-Muslim
majority do not experience similar price effects during the holy
month. We also show that the results are unrelated to well-docu-
mented calendar anomalies such as the January effect or the
Turn-of-the-Week effect. This is perhaps not surprising, as Ram-
adan does not coincide systematically with a xed day or season in
the Gregorian calendar.
The remainder of this paper is organized as follows: Section 2
presents a discussion of the literature on the health benets result-
ing from fasting and the evidence of the market effects of religious
festivities. Section 3 outlines the data set and the data collection
procedures. Section 4 discusses the results, while Section 5 exam-
ines the robustness of these results. The last section concludes the
paper.
2. Related studies of fasting and religious festivities
This section consists of two parts. The rst part is a review of
the clinical research into the health effects of fasting. In the second
part, we review the evidence on the effects of Ramadan on the Is-
lamic stock markets, as well as the effects of other holidays on
markets elsewhere around the world.
2.1. The Ramadan fast and its health benets
Abstaining from ingesting food and liquids between sunrise and
sunset is an integral part of Ramadan. With a suhoor pre-dawn
meal and an iftar supper the consumption becomes entirely noc-
turnal. The Ramadan-type fasting
3
can have a wide range of bene-
cial effects for health. Although in the initial stage of fasting
individuals may experience a feeling of discomfort, this symptom
typically abates after the second day of the fast. Proponents of natu-
ral medicine argue that it arises as a by-product of the detoxication
process, in which the toxins previously stored in fatty tissues are ex-
pelled from the body (Fuhrman, 1998). Fasting also provides a re-
spite to the digestive tract and allows for the elimination of
superuous tissue. A number of clinical studies have reported a
reduction of body weight and waist circumference in subjects under
controlled fasting conditions (Bck et al., 1978; Perk et al., 2001;
Saleh et al., 2005; Bouhlel et al., 2008). The ramications of this
observation for the cardiovascular system and oxygenation are im-
mense. Moreover, Saleh et al. (2005) report a signicant reduction
in total cholesterol and LDL-cholesterol in male subjects, as well as
a decrease in the atherogenic index for the whole sample during
the fast of Ramadan.
Periodic food deprivation can also alleviate the symptoms of
psychosomatic and mental diseases, as well as reduce the tension
and anxiety of patients (see Yamamoto et al., 1979; Yamamoto,
1980). In view of these ndings, it is perhaps unsurprising that
the number of reported suicide attempts drops during the month
of Ramadan (Daradkeh, 1992). Furthermore, many of those who
fast tend to experience mild states of euphoria. As the human body
enters into a fasting mode, its stores of glucose are progressively
used up. Ketone bodies produced in the liver start to supplant glu-
cose as an energy source for the brain (Knerr and Pearl, 2008).
Interestingly, one of the ketone bodies, b-hydroxybutyrate, is an
isomer of GHB (c-hydroxybutyrate), which is known to the medi-
cal profession as an antidepressant and is also used illegally by
some people as a mood-enhancing drug. Noting the structural sim-
ilarities between the abovementioned ketone body and GHB,
Brown (2007) hypothesizes that they may have comparable effects
on the brain, which would explain the diet-induced euphoria.
2.2. Previous studies of religious festivities
Numerous studies have examined security returns for the pres-
ence of recurring seasonal patterns in the Gregorian calendar.
Among other ndings, these studies report persistent anomalous
returns around the turn of the week, the turn of the month, and
the turn of the year (for example, see Lakonishok and Smidt,
1988). However, there are only a few studies that investigate the
market effects of non-secular festivities such as Ramadan. In this
section we present a brief overview of the more relevant research.
Seyyed et al. (2005) examine the behavior of the Saudi Arabian
stock market in Ramadan during the period of 19852000. They
nd no signicant change in mean return but a noticeable decline
3
The term fasting, as used in this paper, describes a fasting-refeeding cycle with
two meals being consumed daily. Under no circumstances do we try to claim that
total abstention from food leads to health improvements in the long-run, nor that it
should be recommended as a treatment for serious medical conditions.
836 J. Biakowski et al. / Journal of Banking & Finance 36 (2012) 835845
in volatility. Ours is a more comprehensive study, which includes a
cross-section of countries observed over a relatively long period.
While our results for Saudi Arabia are not dissimilar to those of
Seyyed et al. (2005), we document that most of the other Islamic
markets have historically experienced large stock price increases
during the holy month. Importantly, we are the rst to document
robustly the existence of this stock market anomaly. Finally, our
paper builds on the results already established in different elds
of science such as medicine, psychology and sociology. The nd-
ings presented here, therefore, cannot be subject to the common
data snooping critique that may be relevant in the context of other
studies (see Crack, 1999; Subrahmanyam, 2007).
A related stream of research is the literature on the effects of
religious holidays and other national festivities. Lakonishok and
Smidt (1988), as well as Ariel (1990) and Cadsby and Ratner
(1992), report that independently of other calendar-based anoma-
lies stock returns tend to increase prior to public holidays during
which markets are closed. These holidays include Christmas and
Good Friday, two important religious holidays.
In a similar study, Frieder and Subrahmanyam (2004) test the
effect of Jewish sentiment on the US equity market by examining
return and volume around major Jewish High Holy Days on which
the stock market remains open. This study nds that stock returns
are signicantly up on Rosh HaShanah (a festive day, commonly
known as the Jewish New Year) and the prior two days, but signif-
icantly down on Yom Kippur (a somber day) and the day after. For
both Holy Days, Frieder and Subrahmanyam (2004) report a de-
cline in the volume of trading.
4
Frieder and Subrahmanyam
(2004) attribute their results to the sentiment of Jewish investors
and their trades around these holidays.
Like Rosh HaShanah, Ramadan is an event capable of inuenc-
ing the moods and decisions of the adherents. During Ramadan
participating Muslims seek a closer relationship with Allah and fol-
low a set of prescribed standards of behavior intended to make
them become better Muslims and more responsible members of
society, which can improve their feelings of self-worth. Thus, we
expect Ramadan to produce an upbeat sentiment, overcondence
and a greater willingness to accept risk by Muslim investors.
3. Data
Ramadan is the ninth month in the Islamic (or Hijri) calendar,
which is based on the motion of the moon. Obtaining historical
information on the duration of Ramadan is a demanding task, as
the holy month does not have a xed date in the Gregorian calen-
dar. The Islamic calendar is used mainly for observance of religious
holidays and functions, whereas the Gregorian calendar is used for
business and government. The 12 months derived from the lunar
cycle are separated by the appearance of the new moon and the
number of days in a month averages between 29 and 30 days, mak-
ing the Islamic year approximately 11 days shorter than the Grego-
rian one. In this paper the all-important event dates are not
approximated, but rather derived from precise astronomical
calculations.
If a new moon sighting occurs before sunset, Ramadan will start
the following day. The subsequent new moon breaks the fast and
the celebrations related to the Eid al-Fitr festival follow. The
technical difculty that arises at this stage is that both the visibility
of the moon and the sunset time are dependent not only on the
date, but also on the geographical location of the observer. In
performing the calculations, we have explicitly assumed that the
celestial bodies are observed from the perspective of individuals
residing in the capital cities of the countries covered in our study.
The latitudes and longitudes of these cities were obtained from the
National Geospatial-Intelligence Agency. These coordinates were
then imputed into an applet provided courtesy of Astronomical
Applications Department at the US Naval Observatory in order to
obtain information on the lunar phases and sunset times at each
location. Implementation of this procedure allowed the determina-
tion of the exact starting and ending dates of the holy month.
Our sample includes countries for which stock market index
data were available from Morgan Stanley Capital International
(MSCI) in Datastream and in which the proportion of population
professing Muslim faith exceeded 50%. The data on population
and religious fractionalization was sourced from the Central Intel-
ligence Agency (CIA)s World Factbook (2009). Data on countries for
which the CIA did not record the exact proportion of Muslims in
the society was extracted from the Association of Religion Data Ar-
chives. As can be seen from Table 1, the nal sample comprises 14
countries inhabited by nearly 695 million people. For each country,
the date on which daily observations on the MSCI index rst be-
came available determines the number of events. It needs to be
mentioned that the event study analysis performed in the next sec-
tion requires at least 200 observations before the rst Ramadan
starting date, so that a benchmark model for the returns can be
estimated. Guided by these considerations, we arrive at a sample
of 129 events.
Although the Muslim community is dominant in the societies
considered, on average accounting for 90.9% of the total popula-
tion, any foreign participation in the stock markets could poten-
tially diminish the importance of this gure in the context of our
study. In order to gauge the extent of non-Muslim foreign owner-
ship, we have consulted the OSIRIS database published by Bureau
van Dijk. The direct ownership data was extracted by manually
matching the SEDOL codes of all of the MSCI index constituent
companies appearing on the MSCI Frontier and Emerging Markets
lists at the end of our sample period with the OSIRIS entries. To en-
sure that the matching procedure is accurate, the codes were cross-
checked with the Datastream database. In our calculations we have
assumed that any investor residing/headquartered outside the 14
countries listed in Table 1 is a non-Muslim. For the sample as a
whole, the capitalization-weighted average of non-Muslim foreign
direct ownership computed across companies included in MSCI
country indices is as low as 9.1%. Consequently, it may be argued
that the followers of Islam hold a majority stake in the local stock
markets
5
and that their investment decisions play a key role in the
price formation process.
4. Results
4.1. Stock market performance during Ramadan
Fig. 1 plots the average returns during the holy month and the
rest of the year. For the purposes of comparison, these continu-
ously compounded returns were denominated in US dollars and
annualized. Visual inspection of the bar chart reveals that the dif-
ferences in means across markets are quite striking, with 11 out of
14 countries having higher average returns during Ramadan. In
aggregate, the mean annualized dollar-denominated return real-
ized by investors during the holy month was 38.09%, compared
to a rather modest gain of 4.32% throughout the rest of the year.
The results reported for the three countries that did not experience
4
Frieder and Subrahmanyam (2004) also analyze market data around the Christian
Feast Day of St. Patrick and nd results similar to those they report for Rosh
HaShanah.
5
The fact that the reported non-Muslim foreign ownership is low does not
necessarily imply that the volume of trading generated by non-Muslims is negligible.
However, due to constraints on data availability, we are unable to provide any trading
volume statistics by religious afliation of traders.
J. Biakowski et al. / Journal of Banking & Finance 36 (2012) 835845 837
this apparent Ramadan effect have to be interpreted with great
caution. The inferences for both Bahrain and Saudi Arabia are
based on a limited number of observations, while much of the
hardship experienced by Indonesia during the Asian Crisis coin-
cided with Ramadan.
Although the MSCI indices are available for a large cross-section
of countries, one of the concerns which can be raised is paucity of
data for Saudi Arabia, a country crucial to our investigation. Con-
sidering that Saudi Arabia is the country of birth of Prophet
Muhammad (PBUH), a place from which Islam originated and
which attracts large numbers of pilgrims, further examination is
required. In an attempt to circumvent this data problem, we down-
loaded and analyzed market capitalization-weighted indices com-
piled by Standard & Poors from Datastream for all countries that
had daily series of sufcient length.
6
This alternative dataset com-
prised 8 national stock markets and 59 Ramadan events (13 events
for Jordan, 12 for Pakistan, 11 for Saudi Arabia, 7 for Bahrain and
Oman, and 3 for United Arab Emirates, Qatar and Kuwait).
7
The evi-
dence of the Ramadan effect was compelling and observed in all of
the sample countries. Inclusion of additional observations for Bah-
rain and Saudi Arabia strengthened the results in favor of the anom-
aly, with the annualized returns averaging 40.77% during the holy
month compared to 11.87% during all of the remaining lunar
months. The null hypothesis of mean return equality was strongly
rejected (t-statistic = 3.65).
8
Overall, this preliminary data analysis
reveals a rather curious anomaly and underscores the need for fur-
ther examination.
To this end, two versions of an event study analysis were per-
formed. In the rst version, the benchmark returns were generated
by a constant-mean-return model, whereas the second event study
considered the observed returns in relation to the predictions of
the market model.
9
For the sake of brevity, we report only estimates
based on the market model and note that the constant mean ap-
proach produced slightly higher cumulative abnormal returns. In
our calculations, a value-weighted MSCI World Index comprising
23 industrialized countries was taken as a proxy for the market port-
folio. None of the countries used to make up this index meet our def-
inition of being predominantly Muslim, which implies that the
selected benchmark remains reasonably event-independent. In order
to ensure that the estimation window does not include the preced-
ing years Ramadan, its length was restricted to 200 trading days.
The abnormal returns were dened conventionally as the returns
in excess of what the investor would expect in absence of the event.
These returns were subsequently averaged over all events and
summed over time to form cumulative abnormal returns (CARs).
The statistical signicance of CARs was veried using a para-
metric t-test similar to that of Kothari and Warner (2007). Notably,
this testing approach disposes of the homoscedasticity assumption
and is able to accommodate event-induced changes in volatility.
The test statistic can be expressed as:
tCARn
1
; n
2

CARn
1
; n
2

rAR

n
2
n
1
1
p 1
where n
1
and n
2
are, respectively, the beginning and end dates of a
given event window and r(AR) is the standard deviation of mean
abnormal returns computed in the time-series dimension over the
(n
1
, n
2
) period. Furthermore, the evidence is supplemented by the
results of a sign test, which does not make any assumptions regard-
ing the distribution of underlying returns. Following Banerjee and
Eckard (2001), we dene the z-statistic as follows:
z
p 0:5

p1 p=N
p 2
where p is the fraction of positive CARs in our sample and N is the
number of events.
Table 1
Sample composition. This table lists the markets included in our analysis with the corresponding number of events. In order to be included in the sample, a country needed to
have an MSCI stock market index in the DataStream database and a proportion of Muslims in total population exceeding 50%. The data on population and religious
fractionalization has been obtained from the Central Intelligence Agency (CIA) World Factbook (2009). In cases where the CIA did not record the exact proportion of Muslims in the
society, the data was sourced from the Association of Religion Data Archives. The fourth column reports information on the direct foreign ownership by investors residing/
headquartered outside the 14 countries listed in the table below. These statistics are capitalization-weighted, computed across all MSCI Index constituents and based on the
ownership data available in the OSIRIS database. Since the dates of the holy month of Ramadan are determined by a lunar cycle observed at a particular geographical location, we
collected the coordinates of capital cities from the National Geospatial-Intelligence Agency. These coordinates are used subsequently to calculate the moons phases from which
the starting and ending dates of events are derived. MSCI Index Start Date is the date at which daily observations on a particular index become available.
Country Population Percent
muslim (%)
Non-muslim
foreign
ownership (%)
Capital city Latitude Longitude MSCI index
start date
Number of
observations
Bahrain 727,785 81.2 9.0 Manama 26 14 10 N 050 34 59 E 31-May-05 2
Egypt 83,082,869 90.0 28.2 Cairo 30 03 00 N 031 15 00 E 30-December-94 13
Indonesia 240,271,522 86.1 21.2 Jakarta 06 10 28 S 106 49 46 E 31-December-87 20
Jordan 6342,948 92.0 10.0 Amman 31 57 00 N 035 56 00 E 31-December-87 20
Kuwait 2691,158 85.0 3.5 Kuwait 29 22 11 N 047 58 42 E 31-May-05 2
Malaysia 25,715,819 60.4 14.9 Kuala Lumpur 03 10 00 N 101 42 00 E 31-December-87 20
Morocco 34,859,364 98.7 28.5 Rabat 34 01 12 N 006 49 48 W 30-December-94 13
Oman 3418,085 89.2 3.6 Muscat 23 36 48 N 058 35 36 E 31-May-05 2
Pakistan 176,242,949 95.0 3.7 Islamabad 33 42 00 N 073 10 00 E 31-December-92 15
Qatar 833,285 77.5 12.2 Doha 25 17 12 N 051 32 00 E 31-May-05 2
Saudi Arabia 28,686,633 100 1.3 Riyadh 24 38 27 N 046 46 22 E 31-May-05 2
Tunisia 10,486,339 98 17.1 Tunis 36 48 10 N 010 10 47 E 31-May-04 3
Turkey 76,805,524 99.8 16.2 Ankara 39 55 38 N 032 51 52 E 31-May-94 13
United Arab Emirates 4798,491 96.0 1.9 Abu Dhabi 24 28 00 N 054 22 00 E 31-May-05 2
Total 694,962,771 90.9 9.1 129
6
A series was considered to be of sufcient length if it spanned back at least to
01/01/2006.
7
We did not merge the S&P and MSCI indices into one sample because of
methodological differences in index construction. In selecting index constituents, S&P
uses xed size and liquidity cutoff points. MSCI, on the other hand, relies on cutoff
points derived from rankings. Furthermore, MSCI indices are designed to cover 99% of
free oat, whereas S&P targets 80% of capitalization. The frequency of index
reconstitutions is also dissimilar for these two data providers.
8
More detailed results obtained for the S&P indices are available from the authors
upon request.
9
A detailed description of both approaches can be found in Campbell, Lo and
MacKinlay (1997).
838 J. Biakowski et al. / Journal of Banking & Finance 36 (2012) 835845
Depending on the motion of the moon, the holy month of
Ramadan lasts either 29 or 30 days. This translates into 2122 days
of trading, or equivalently into a (0, 20) or (0, 21) event window.
According to the results reported in Table 2, the cumulative abnor-
mal returns during the considered period are strongly statistically
signicant. The CARs are within the 2.442.48% range and these
estimates can increase to as much as 4.99% once we focus on the
(10, 35) event window and use the constant-mean-return model.
The intriguing pattern revealed by the plot of CARs in Fig. 2
warrants further elaboration. First, the abnormal growth in stock
prices during the rst 5 days of the holy month is negligible. This
is perhaps not entirely surprising, as a fraction of people may
experience some discomfort in the initial stages of fasting arising
as a by-product of the detoxication process and the changed
dietary regimen. These symptoms, however, subside quickly and
the subsequent increased production of ketone bodies will
improve the mood of a representative investor. What also becomes
apparent when examining Fig. 2 is that stock prices continue to rise
for a number of days after the fast is complete. To a certain extent
this may be a reection of the sense of accomplishment that comes
with achieving spiritually meaningful objectives, a sense that
permeates the whole society. These positive feelings as well as the
communal spirit are likely to be reinforced by the three-day-long
Eid al-Fitr festival commencing upon the completion of the Ram-
adan fast. Muslims who partake in the festivities tend to spend a
greater portion of their time in a circle of family and friends, and
on religious services that can reinforce an optimistic view of the
world.
An issue that may arise when evaluating the statistical signi-
cance within a standard event-study framework is the contempo-
raneous correlation of abnormal returns in the cross-sectional
dimension. Since the periods of the holy month tend to partially
overlap across countries, one could suspect that such correlation
may be present in our data. A number of authors have argued that
the problem of event clustering can be addressed by aggregating
returns into a portfolio and regressing the resultant portfolio re-
turns against an event dummy variable (Binder, 1985, 1998;
Bartholdy et al., 2004). Although this approach has the potential
to remedy the complications arising from cross-correlation of
returns, it may also result in a reduction in power (Bernard, 1987).
The regression reported below links the returns on an equally-
weighted portfolio constructed from the stock market indices of
the 14 predominantly Muslim nations in our sample with the
returns on the MSCI World Index and a Ramadan dummy variable.
Since the dates of the holy month differ slightly across countries,
the dummy starts recording a value of one with the rst
Table 2
Cumulative abnormal returns. This table reports cumulative abnormal returns around the holy month of Ramadan along with the results of statistical
signicance tests. The cumulative abnormal returns are expressed in US dollars and have been computed based on a sample of 129 events. The abnormal
returns are expressed as deviations from the predictions of the market model. The third and fourth columns present t-statistics with the corresponding p-
values for the hypothesis that the cumulative abnormal returns equal zero. The last three columns in the table report the results of the non-parametric sign
test. Under the null hypothesis that the proportion of positive cumulative abnormal returns in the whole sample equals 50% the z-statistic is normally
distributed with mean zero and standard deviation of one.
Event
window
CAR (%) t-stat p-value Proportion of
positive CARs
z-stat p-value
(0, 5) 0.0345 0.0867 0.9310 0.5116 0.2642 0.7916
(0, 10) 0.6054 0.3757 0.7078 0.5891 2.0580 0.0396
(0, 20) 2.4798 2.5376 0.0124 0.6357 3.2017 0.0014
(0, 21) 2.4381 2.4651 0.0150 0.6434 3.4005 0.0007
(0, 30) 3.5754 3.2649 0.0014 0.6667 4.0156 0.0001
(10, 20) 3.6306 3.4483 0.0008 0.6357 3.2017 0.0014
(10, 21) 3.5962 3.3826 0.0010 0.6357 3.2017 0.0014
(10, 35) 4.1363 3.2200 0.0016 0.6047 2.4311 0.0151
Fig. 1. Annualized raw stock returns for MSCI stock market indices during Ramadan and non-Ramadan periods. Note: This gure depicts annualized continuously
compounded returns in 14 predominantly Muslim countries during the holy month of Ramadan and throughout the rest of the year. All returns are computed using MSCI
indices denominated in US dollars.
J. Biakowski et al. / Journal of Banking & Finance 36 (2012) 835845 839
commencement of Ramadan and reverts back to zero when the holy
month is ended in all nations. The t-statistics based on the White
(1980) corrected standard errors are given in parentheses.
Return Portfolio
t
0:01
1:09
0:11
3:15
Ramadan
t
0:22
10:43
Return World
t
e
t
Adj: R
2
4:04%
3
Since the dependent variable is expressed in percentage terms, the
value of the coefcient on Ramadan implies that, ceteris paribus, re-
turns are 11 basis points per day higher during the holy month. The
p-value associated with the null hypothesis that this coefcient is
equal tozerois as lowas 0.0016, collaboratingour earlier conclusions.
An alternative methodology capable of dealing with the issue of
overlapping events is a variant of the calendar-time portfolio ap-
proach advocated by Jaffe (1974), Fama (1998) and Mitchell and
Stafford (2000). It is implemented here by applying the event study
analysis to the equally-weighted portfolio constructed from na-
tional indices. Since the estimate of portfolio variance automati-
cally takes account of any potential cross-country dependence,
the parametric test statistic for CARs is unlikely to be overstated
(Mitchell and Stafford, 2000). However, the obvious deciency of
this approach is the reduction in sample size from 129 to 19. Not-
withstanding the smaller number of observations, we nd that the
CARs retain their statistical signicance in this model setting. The
t-statistics for the (0, 20) event window are 3.02 for the market
model-adjusted cumulative abnormal returns and 3.43 for the con-
stant-mean-model. Both of the corresponding p-values fall com-
fortably below the 1% level. This conrms that the seasonal effect
tested in this paper is unlikely to be a mere by-product of the spe-
cic testing methodology adopted.
For our explanations to be plausible and consistent, it also
needs to be shown that the effect induced by Ramadan is weaker
in countries where the proportion of population adhering to the Is-
lamic faith is smaller. In order to verify this hypothesis, we col-
lected data on all 67 countries for which the MSCI indices were
available in Datastream. These nations were subsequently divided
into three subgroups according to the religious fractionalization of
the society. Group A comprises the 14 nations in which the fraction
of Muslims in total population exceeded 50%, Group B includes an-
other 14 countries in which this fraction fell between 5% and 50%,
and Group C consists of the 39 nations which did not meet the 5%
threshold. A portfolio was constructed for each of the groups by
equally weighting the dollar-denominated returns on the national
stock market indices.
These portfolio returns were then linked to the Ramadan zero-
one dummy variable and additional controls using a Multivariate
Regression Model. This methodological approach is similar to that
used in Binder (1985, 1998), Wagster (1996) and Bartholdy et al.
(2004) and employs the seemingly unrelated regression (SUR)
technique introduced by Zellner (1962) to estimate a system of
equations. The results of the SUR estimation for our 3 country-
groups are reported in Table 3 and appear to accord with the basic
intuitive predictions. The effects of Ramadan materialize only
when the society chooses to participate in this religious experience
collectively. Unsurprisingly, the joint hypothesis that the coef-
cients on the Ramadan dummy in the regressions for groups B
and C are equal to zero cannot be rejected.
10
4.2. Stock market volatility during the holy month
Although the cumulative abnormal returns are both statistically
and economically signicant, one has to ensure that they are not
merely a compensation for increased risk during the Ramadan per-
iod. In order to detect any potential changes in the riskiness of the
stock markets an examination of return volatility was undertaken.
Fig. 3 depicts a geometric average of annualized standard devia-
tions calculated across all of the 129 events. The estimation of vol-
atilities follows a moving-window approach, in the sense that for
any given day in the event window the standard deviations were
measured over the previous 20 trading days. This means that val-
ues recorded for days 1921 are representative of the index vola-
tility observed during the holy month. Quite remarkably, these
days coincide with a notable dip on the graph. One can conse-
quently conclude that the observed Ramadan effect cannot be
easily rationalized within the market efciency framework and
ought to be regarded as anomalous.
Table 4 presents additional results regarding the arithmetic
averages of unconditional standard deviations disaggregated by
country. With the exception of Turkey, all of the countries experi-
enced a drop in index volatility during Ramadan. The statistical sig-
nicance of this phenomenon was tested using a Wilcoxon signed-
ranks test (nonparametric), which takes into account both the
direction and the relative magnitude of changes (Daniel, 1978;
Kanji, 1993). Under the null hypothesis of constant variance the
test statistic follows a standard normal distribution. The null
hypothesis is, however, convincingly rejected at the conventional
signicance levels. One could explain this result by referring to
the literature on human psychology. Ellision et al. (2009) have
demonstrated empirically that religious beliefs and high degrees
of social integration positively inuence individuals feeling of
tranquility. It is therefore plausible that as many investors reach
the state of inner peace and calmness, the behavior of stock prices
changes to reect this.
4.3. Ramadan and stock market liquidity
Finally, it needs to be noted that previous work has reported a
strong negative relationship between liquidity and expected ex-
cess returns (see Amihud and Mendelson (1986, 1991)).
Consequently, further analysis is needed to assess whether the
abnormal returns observed during the month of Ramadan are not
a manifestation of the illiquidity premium. As described by Ami-
hud (2002), liquidity is a rather elusive concept and a wide range
of proxies can be used to gauge it.
Guided primarily by data availability considerations, we have
adopted four different liquidity measures calculated using
Fig. 2. Cumulative abnormal returns during the month of Ramadan. Note: The
gure above plots the cumulative abnormal returns calculated using the market
model approach. The total sample consists of 129 events in 14 predominantly
Muslim nations. Day 0 on the graphs marks the beginning of Ramadan, which
typically lasts around 2122 trading days.
10
The relevant F-statistic is 1.01 for Panel A of Table 6 (p-value = 0.3628) and 0.51
for Panel B (p-value = 0.5996).
840 J. Biakowski et al. / Journal of Banking & Finance 36 (2012) 835845
company-level data. Two of the measures have been proposed by
Bekaert et al. (2007) and are particularly suitable in the context
of emerging and frontier markets where complete time series on
bid-ask spreads are difcult to obtain. The rst gauge, namely
the average proportion of days with zero rm returns, takes high
values when the informational signals do not outweigh the
transaction costs. Their second measure, called the price pressure
index, takes into account both the frequency of companies zero re-
turn days and their persistence (see Bekaert et al. (2007)). It assigns
a higher illiquidity score whenever the zero return days occur
consecutively, compared to a case where periods of illiquidity
and active trading are interchangeable. This is because the price
pressure build-up is greater in the former case. In calculating these
two constructs, we have focused on returns denominated in local
currency in order to avoid confusion between liquidity of stocks
and liquidity of the foreign exchange market. Furthermore, we
examined the average proportion of company-days with zero
volume. This statistic constitutes the conceptual core of the liquid-
ity measure introduced by Liu (2006). Last but not least, a turnover
ratio, dened as the total value of trading volume divided by the
total market capitalization, was calculated and analyzed.
11
Our def-
inition parallels that employed by Haugen and Baker (1996) and
Bhattacharya and Daouk (2002).
In order to nd the data required for the construction of the
abovementioned liquidity measures, Datastream was searched
for rm-level information. For a company to be included in our
sample it had to be listed in one of the stock markets studied here
Table 3
Comparing Muslim and non-Muslim Countries: seemingly unrelated regressions. All of the countries for which MSCI
index data was available were divided into three groups depending on the proportion of population professing the
Muslim faith. Group A comprises 14 nations in which Muslims constitute a majority; Group B includes 14 countries
where the fraction of Islam adherents in the society falls between 5% and 50%; and Group C consists of 39 nations in
which this fraction is below 5%. Within each of the groups, a portfolio was constructed by equally weighting the dollar-
denominated returns on the national MSCI indices. The daily returns on the resultant three portfolios were
simultaneously regressed on a range of explanatory variables using the Seemingly Unrelated Regression (SUR)
methodology. Ramadan is a dichotomous variable for the Muslim holy month dened here according to the motions of
the moon observed from the perspective of Mecca. Halloween takes the value of 1 during the NovemberApril period
and 0 otherwise. January and Monday are indicator variables for the rst month of the year and rst day of the week,
respectively. Return_World is the continuously compounded return on the value-weighted MSCI World Index.
Coefcient standard errors are reported in parentheses.
Group A Group B Group C
Panel A. Inuence of Ramadan in three different groups of countries
Intercept 0.0046 0.0158
**
0.0125
**
(0.0062) (0.0067) (0.0062)
Ramadan 0.0594
***
0.0170 0.0301
(0.0211) (0.0229) (0.0212)
Return_World 0.0879
***
0.3230
***
0.5785
***
(0.0077) (0.0083) (0.0077)
Return_World_Lag 0.1090
***
0.1712
***
0.1912
***
(0.0077) (0.0083) (0.0077)
System weighted R-square 0.3344
Panel B. Inuence of Ramadan after controlling for other seasonal anomalies
Intercept 0.0074 0.0095 0.0086
(0.0088) (0.0096) (0.0089)
Ramadan 0.0519
**
0.0077 0.0212
(0.0214) (0.0233) (0.0215)
Halloween 0.0127 0.0246
*
0.0211
*
(0.0127) (0.0138) (0.0127)
January 0.0566
**
0.0160 0.0311
(0.0222) (0.0242) (0.0223)
Monday 0.0664
***
0.0320
**
0.0419
***
(0.0148) (0.0161) (0.0149)
Return_World 0.0875
***
0.3226
***
0.5780
***
(0.0076) (0.0083) (0.0077)
Return_World_Lag 0.1091
***
0.1710
***
0.1911
***
(0.0076) (0.0083) (0.0077)
System Weighted R-square 0.3360
*
Statistical signicance at 10%.
**
Statistical signicance at 5%.
***
Statistical signicance at 1%.
Fig. 3. Rolling average volatility. Note: This gure plots a geometric average of the
annualized standard deviations of returns computed over 129 events. The volatility
is measured over the last 20 trading days relative to a particular day in the event
window. Day 0 denotes the start of Ramadan and days 2021 on the graph above
mark its end.
11
The calculation of the turnover ratio was based on the aggregate dollar-
denominated daily trading volumes of all sample companies divided by their
aggregate market capitalizations.
J. Biakowski et al. / Journal of Banking & Finance 36 (2012) 835845 841
and it had to have complete data on trading volume and capitaliza-
tion throughout the entire sample period. The latter criterion
ensured that the sample composition remained unaltered through-
out the entire span of the analysis. In total, 364 companies met the
sample inclusion criteria.
12
The values of the four liquidity
measures computed based on rm-level information are reported
in Table 5. The rst three of them do not record any statistically
signicant changes during the Ramadan period. Daily Turnover Ratio,
on the other hand, exhibits a notable jump with a corresponding
t-statistic of 2.66. While further research is needed in order to assess
the impact of Ramadan on stock market liquidity, we cannot nd any
evidence of a statistically signicant decrease.
5. Robustness checks
Although the prevalent view among Muslims is that the holy
month starts when the birth of a new moon is observed locally, a
small minority focuses on the visibility of the crescent from the
perspective of Mecca and fasts accordingly. The Islamic Society of
North America, for instance, disseminates information about Ram-
adan dates, announced by the Fiqh Council of North America, based
solely on the Saudi lunar cycle. To verify the robustness of results,
the event-study analysis was repeated under the assumption that
the dates of the holy month in all countries coincide with the sight-
ing of the new moon in Mecca. The cumulative abnormal returns
retain their statistical signicance and deviate only by a couple
of basis points from the estimates of our baseline model. Conse-
quently, a change in the interpretation of the sighting does not al-
ter the main conclusions of this paper.
Fasting can become particularly demanding on long summer
days when the break between the meals is protracted and the risk
of dehydration is higher. One could therefore speculate that such
conditions would be less conducive to good mood. This contention
was investigated by looking at the relationship between the cumu-
lative abnormal returns and the length of the day from sunrise to
sunset. The day length was measured in minutes on Ramadans
rst day of trading. The analysis revealed that there is indeed a
negative relationship between the cumulative abnormal returns
and the length of the fast day, but the correlation coefcients were
negligible and statistically insignicant. This nding attests to the
fact that the Ramadan-induced abnormal returns are relatively
insensitive to the changes in duration of the within-day fasting
period.
13
Before it can be claimed that the Ramadan effect is an anomaly
in its own right, it needs to be demonstrated that it is truly inde-
pendent of other calendar effects reported in the literature. Among
these, the turn-of-the-year seasonality has likely received the most
attention from the academic community. The observation that the
U.S. stock market returns are elevated in January was rst made by
Rozeff and Kinney (1976) and later conrmed in a study using an
international sample by Gultekin and Gultekin (1983). Sias and
Starks (1997) argued that this evidence is consistent with tax-
loss-selling, while Keim (1983) viewed it through the prism of
the size effect. In what follows, we exclude all holy months com-
mencing in January from our sample, thereby reducing the number
of relevant events to 115. The plot presented in Panel A of Fig. 4
conrms that our results remain robust even after the month of
January was discarded. The annualized return during an average
ninth month of the Muslim calendar, which did not start in Janu-
ary, is 30.38%. This gure is signicantly higher (at 5% level) than
the returns realized by investors throughout the rest of the year.
Another seasonality we focus on is what is commonly labeled
the Halloween effect. Bouman and Jacobsen (2002) conclude that
it is optimal for investors to pursue a Sell in May and go away
strategy, staying in cash until the end of October and then invest-
ing in stocks, as the average stock returns during the November
April period are much higher compared to the other half-year. Con-
sequently, the second sub-sample we construct comprises all
months of Ramadan that did not start between November and
April. This condition was met by 50 events and the ndings are de-
picted in the Panel B of Fig. 4. Accounting for the Halloween anom-
aly only strengthens our conclusions with regard to the existence
of a Ramadan effect. In this case, the average annualized returns
during the event period amounted to a staggering 51.78% (corre-
sponding t-statistic = 4.44).
The seasonal patterns in stock returns were also controlled for
in a multivariate framework. In addition to the January and Hal-
loween anomalies, Table 6 considers the well-documented Mon-
day effect (Gibbons and Hess, 1981; Keim and Stambaugh, 1984).
Panel A of the table reports the results of pooled OLS estimation,
whereas Panel B employs a xed effect panel approach. The latter
methodology accommodates the time-invariant heterogeneity
across countries and is relatively robust to any possible omitted
variable biases (Chamberlain, 1978). The results reveal that the cal-
endar anomalies observed in the more developed capital markets
can also be documented for the counties in our sample. Most
importantly, the Ramadan binary variable is statistically signicant
in all of the specications and the impact of the Muslim holy
month on the rst moment of return distribution appears to be
unrelated to changes in the value of world market portfolio.
We also investigate whether the anomaly has persisted over
time. Had the effect been discovered by market participants prior
to the end of our sample, they would have adjusted their trading
strategies to take advantage of it. In doing so, they would have
made the markets more efcient. We consider returns during
two periods: 19892000 and 20012007. This split generates
two sub-samples with comparable number of observations. The
gap in annualized returns between Ramadan and the rest of the
year actually increased from 31.98% to 35.59% between the rst
and second intervals. Consequently, the results do not suggest that
Table 4
Return volatility during Ramadan and the rest of the year. This table juxtaposes the
arithmetic averages of annualized unconditional standard deviations of returns
during the holy month of Ramadan and all the remaining months of the Islamic
calendar. The statistical signicance of the average difference between these two
estimates is tested using a Wilcoxon signed-ranks test. In large samples, the Wilcoxon
test statistic has approximately a standard normal distribution under the null.
1 Ramadan (%) Rest of the year (%) Difference (%)
Bahrain 8.5432 16.7052 8.1620
Egypt 21.2382 22.5039 1.2657
Indonesia 31.0700 32.9271 1.8570
Jordan 10.4508 16.0438 5.5930
Kuwait 11.1855 21.5666 10.3811
Malaysia 18.8060 22.8885 4.0825
Morocco 12.1587 13.0228 0.8641
Oman 10.7488 14.5779 3.8290
Pakistan 24.2362 29.7812 5.5450
Qatar 20.1577 23.6476 3.4899
Saudi Arabia 18.3109 42.9538 24.6428
Tunisia 9.1412 14.2941 5.1529
Turkey 59.0653 47.7005 11.3649
United Arab Emirates 13.2271 28.6732 15.4461
Total 22.9757 25.6201 2.6444
Wilcoxon test statistic 4.9657
12
It may be argued that Datastream tends to cover primarily high proles stocks.
However, after dividing our sample into two sub-samples of large and small
capitalization stocks, we have discovered that our inferences with regard to liquidity
are similar for both groups. This nding mitigates the concerns regarding our sample
composition.
13
A full version of results related to our robustness checks is available upon request.
842 J. Biakowski et al. / Journal of Banking & Finance 36 (2012) 835845
Islamic stock markets were becoming increasingly efcient over
time.
Lastly, we analyze whether the effect described in this study
arises from the uctuations observed in the foreign exchange
market. We operationalize this analysis by repeating the return
calculations for MSCI indices, this time denominated in local cur-
rency, and contrasting the results with our initial ndings. As
expected, the conclusions change substantially in favor of the
anomaly for Indonesia, where the occurrence of holy month coin-
cided with the Asian crisis. On the other hand, the change of
Table 5
Liquidity during the holy month. The table below presents values of four liquidity measures during Ramadan and the rest of the year. Calculations are based on 364 companies
listed in countries where Muslims account for the majority of population. The sample is restricted to stocks for which complete data was available in Datastream. The rst
measure labeled Proportion of Zero Return Days was designed in the spirit of Bekaert et al. (2007). During the relevant period, we record the fraction of days for which the returns
denominated in local currency were equal to zero and apply equally-weighted averaging across all companies. The Price Pressure Measure, rst introduced by Bekaert et al. (2007,
p. 1789), takes into account both the frequency of zero return days and their persistence. Values closer to 1 indicate a less liquid market. In the rst step of computations, we
construct a price pressure index for each of the markets separately, using returns expressed in domestic currencies. The nal statistic is calculated as an average, weighted by the
number of sample companies available in each of the countries. Proportion of Zero Volume Days gauges the fraction of company-days with no trading. Daily Turnover Ratio denotes
the average of total US dollar equivalent of daily trading volume in all stocks divided by their total US dollar market capitalization. The last three columns report the differences in
liquidity measures, the t-statistics for the null hypotheses that the difference is zero, along with the corresponding p-values. Standard errors are given in parentheses.
Ramadan period Rest of the year Difference t-statistic p-value
Proportion of zero return days 0.3603 0.3570 0.0033 0.4518 0.6633
(0.1413) (0.1354)
Price pressure measure 0.4155 0.4202 0.0047 0.8696 0.4209
(0.2271) (0.2520)
Proportion of zero volume days 0.1351 0.1267 0.0084 1.0785 0.2813
(0.1536) (0.1433)
Daily turnover ratio (in%) 0.2960 0.2620 0.0340 2.6577 0.0081
(0.2530) (0.2330)
Fig. 4. Annualized raw returns during Ramadan and non-Ramadan periods in different sub-samples. Note: This gure plots the annualized continuously compounded US
dollar denominated returns on the MSCI indices during the holy month and the rest of the lunar year. Panel A considers all of the 115 events, which did not start in the month
of January. Panel B, on the other hand, reports results for 50 events, which did not commence in the period between November and April.
J. Biakowski et al. / Journal of Banking & Finance 36 (2012) 835845 843
denomination eliminates the existence of the effect in the Tunisian
stock market. Overall, the annualized difference in returns be-
tween the Ramadan and non-Ramadan periods increases from
33.77% when the returns are measured in US dollars to 35.66%
when they are expressed in domestic currency terms. We thus con-
clude that the Ramadan effect is driven primarily by the changing
values of stock prices, rather than the movements of exchange
rates.
6. Conclusions
Motivated by the evidence from positive psychology showing
that religion affects believers mood, happiness and risk-taking
attitude, we investigate the market effect of Ramadan, one of the
most celebrated religious traditions in the world. Using daily equi-
ty return data for 14 predominantly Muslim countries over the
years 19892007, we nd evidence of a strong Ramadan effect.
Specically, over the holy month the mean annualized return is
signicantly higher than the mean return over the rest of the
Islamic year (38.09% vs. 4.32%). For the Ramadan effect to materi-
alize the majority of population needs to be of Muslim faith.
Data from different providers were examined and ten different
testing approaches were used,
14
each time re-conrming the
robustness of the anomaly. Despite these large returns, we nd a sig-
nicant decrease in volatility in all but one of the sample countries.
These results imply a sharp shift in the investors optimism. We
check whether the effect is due to market liquidity, length of the dai-
ly fasting period, currency uctuations and some well-known xed
calendar anomalies, notably the Monday, January and Halloween ef-
fects. None of these factors explain the results. Hence, we nd the re-
sults anomalous and inconsistent with the notion of market
efciency.
We believe that the Ramadan effect documented in this paper
can best be explained by a change in investor psychology. Rama-
dan is a religious month that affects almost every aspect of Mus-
lims lives. Besides fasting and prayers, Ramadan promotes
heightened social awareness and a closer relationship with Allah
and with other fellowMuslims around the world. As a fundamental
shared experience, Ramadan brings about a sense of solidarity
among Muslims, enhances their satisfaction with life and encour-
ages optimistic beliefs. This optimism affects investor sentiment
and decisions leading to the price run-ups we report in this paper.
Our baseline results have been derived from the analysis of
stock indices of 14 countries. Collectively, they reect the decisions
of many investors in the Muslim world. It would be interesting to
further explore the Ramadan effect at the individual company or
business sector level. Some companies may operate more in line
with the core tenets of Islam and may be more socially responsible
than others. Such companies might benet more from the behav-
ioral biases of pious investors during Ramadan. Further evidence
would help us better understand what drives the Ramadan effect
and what the determinants imply for asset valuation. This under-
standing is important given the potential opportunities these
mostly frontier markets provide to the investing public.
The implication of our ndings for investors is obvious. Inves-
tors seeking fast prots in the Muslim world should try to prot
from the fast, buying shares prior to the start of Ramadan and sell-
ing them at the end of the holy month or preferably immediately
after Eid al-Fitr. Of course, there are transactions costs that one
would need to take into account, but any such costs would appear
to pale in comparison to the observed returns. Alternatively, inves-
tors could embark on a more passive strategy by delaying their
sales until the end of the holy month or by accelerating the timing
of purchases ahead of it. We feel however duty-bound to mention
two caveats. First, the fact that an anomaly was present in the data
for decades gives no guarantees that it will also persist in the fu-
ture. Second, the stock markets tend to be excessively volatile
(Shiller, 1981) and seasonal effects, as a general rule, are incapable
of explaining large proportion of this volatility.
Acknowledgements
The authors would like to thank the Institute of Finance Profes-
sionals New Zealand Inc. for awarding this study the best paper
prize in the investment category and MSCI Barra for providing
useful data. The paper has been beneted from the comments of
an anonymous referee, the editor, Warwick Anderson, Henk Berk-
man, Magdalena Biakowska, Glenn Boyle, Stephen Ciccone, Eric
Crampton, Timothy Crack, Mehrun Etebari, Aaron Gilbert, Robin
Table 6
Controlling for other calendar anomalies: Pooled OLS and panel estimations. The
regressions reported in this table examine the statistical signicance of seasonal
regularities in the dollar-denominated returns in 14 predominantly Muslim countries.
Daily continuously compounded returns on the MSCI indices expressed in percentage
terms have been used as a dependent variable in all of the specications. Panel A
assumes an equal intercept across all countries, whereas the regressions reported in
Panel B include country-specic xed effects. To conserve space, the estimates of
xed effects are not reported in the table. Ramadan is a dummy variable for the
Muslim holy month observed according to the local lunar cycle. Halloween takes the
value of 1 during the NovemberApril period and 0 otherwise. January and Monday
are indicator variables for the rst month of the year and rst day of the week,
respectively. Return_World is the continuously compounded return on the value-
weighted MSCI World Index comprising 23 industrialized countries. The regressions
in presented in this table are based on an unbalanced panel. The length of the series
for each of the countries is determined by data availability described in Table 1.
Coefcient standard errors are reported in parentheses.
(1) (2) (3) (4)
Panel A. Pooled OLS estimation
Intercept 0.0038 0.0254 0.0252 0.0018
(0.0116) (0.0156) (0.0156) (0.0165)
Ramadan 0.1042
***
0.0914
**
0.0868
**
0.0882
**
(0.0387) (0.0390) (0.0390) (0.0390)
Halloween 0.0622
***
0.0407
*
0.0405
*
(0.0222) (0.0234) (0.0234)
January 0.1215
***
0.1223
***
(0.0412) (0.0412)
Monday 0.1355
***
(0.0275)
Return_World 0.1900
***
0.1892
***
0.1896
***
0.1891
***
(0.0141) (0.0141) (0.0141) (0.0141)
Return_World_Lag 0.2386
***
0.2376
***
0.2385
***
0.2385
***
(0.0141) (0.0141) (0.0141) (0.0141)
Adjusted R-square 0.0179 0.0181 0.0184 0.0191
Panel B. Fixed effect panel estimation
Ramadan 0.1043
***
0.0916
**
0.0870
**
0.0885
**
(0.0387) (0.0390) (0.0390) (0.0390)
Halloween 0.0614
***
0.0396
*
0.0394
*
(0.0222) (0.0234) (0.0234)
January 0.1229
***
0.1238
***
(0.0412) 0.0412)
Monday 0.1354
***
(0.0275)
Return_World 0.1903
***
0.1895
***
0.1899
***
0.1895
***
(0.0141) (0.0141) (0.0141) (0.0141)
Return_World_Lag 0.2389
***
0.2379
***
0.2388
***
0.2388
***
(0.0141) (0.0141) (0.0141) (0.0141)
R-square 0.0181 0.0183 0.0185 0.0193
*
Statistical signicance at 10%.
**
Statistical signicance at 5%.
***
Statistical signicance at 1%.
14
The statistical and econometric methods employed in this paper included: a
simple test for equality of two mean returns, parametric t-test and non-parametric
z-test in both constant-mean-adjusted and market-model-adjusted event study,
portfolio regressions, portfolio-based event study, pooled OLS regressions, xed effect
panels, and SUR models.
844 J. Biakowski et al. / Journal of Banking & Finance 36 (2012) 835845
Grieves, Ben Jacobsen, Jayant Kale, Brendan Lambe, Warren McNoe,
Philip Meguire, Debra and Bob Reed, participants of the 14th New
Zealand Finance Colloquium, the 2nd Finance and Corporate Gov-
ernance Conference at the La Trobe University, the 2010 Annual
Meeting of the Academy of Behavioral Finance & Economics, and
seminar participants at the University of New Hampshire, Univer-
sity of Otago, University of Canterbury, Queensland University of
Technology and the European University Viadrina. Tomasz Piotr
Wisniewski would like to acknowledge the support of the Univer-
sity of Leicesters sabbatical scheme. The authors retain the sole
responsibility for all remaining errors. An earlier version of this pa-
per has been circulated under the title Piety and Prots: Stock
Market Anomaly during the Muslim Holy Month.
References
Amihud, Y., 2002. Illiquidity and stock returns: cross-section and time series effects.
Journal of Financial Markets 5 (1), 3156.
Amihud, Y., Mendelson, H., 1986. Asset pricing and the bid-ask spread. Journal of
Financial Economics 17 (2), 223249.
Amihud, Y., Mendelson, H., 1991. Liquidity, asset prices and nancial policy.
Financial Analysts Journal 47 (6), 5666.
Ariel, R.A., 1990. High stock returns before holidays: existence and evidence on
possible causes. Journal of Finance 45 (5), 16111626.
Banerjee, A., Eckard, E.W., 2001. Why regulate insider trading? Evidence from the
rst great merger wave (18971903). American Economic Review 91 (5), 1329
1349.
Bartholdy, J., Boyle, G.W., Stover, R.D., 2004. Deposit insurance and the stock
market: evidence from Denmark. European Journal of Finance 10 (6), 567578.
Bekaert, G., Harvey, C.R., Lundblad, C., 2007. Liquidity and expected returns: lessons
from emerging markets. Review of Financial Studies 20 (5), 17831831.
Beit-Hallahmi, B., Argyle, M., 1997. The Psychology of Religious Behaviour. Belief
and Experience. Routledge, London.
Bernard, V.L., 1987. Cross-sectional dependence and problems in inference in
market-based accounting research. Journal of Accounting Research 25 (1), 148.
Bhattacharya, U., Daouk, H., 2002. The world price of insider trading. Journal of
Finance 57 (1), 75108.
Binder, J.J., 1985. On the use of the multivariate regression model in event studies.
Journal of Accounting Research 23 (1), 370383.
Binder, J.J., 1998. The event study methodology since 1969. Review of Quantitative
Finance and Accounting 11 (2), 111137.
Bck, D., Khle, K., Weimann, G., Thomas, W., Mente, F., Schmidt, T., Jaeger, M., 1978.
Prospective studies of the relationship between psychological and social
symptoms with the long-term success of hospital fasting therapy.
Verhandlungen Der Deutschen Gesellschaft Fur Innere Medizin 84, 15651567.
Bouhlel, E., Zaouali, M., Miled, A., Tabka, Z., Bigard, X., Shephard, R., 2008. Ramadan
fasting and the GH/IGF-1 axis of trained men during suboptimal exercise.
Annals of Nutrition and Metabolism 52 (4), 261266.
Bouman, S., Jacobsen, B., 2002. The Halloween indicator, Sell in May and go away:
another puzzle. American Economics Review 92 (5), 16181635.
Brown, A.J., 2007. Low-carb diets, fasting and euphoria: is there a link between
ketosis and c-hydroxybutyrate (GHB)? Medical Hypotheses 68, 268271.
Brown, S., Taylor, K., 2010, Social interaction and stock market participation:
evidence from British Panel Data. IZA Discussion Paper, No. 4886.
Cadsby, C.B., Ratner, M., 1992. Turn-of-month and pre-holiday effects on stock
returns: some international evidence. Journal of Banking and Finance 16 (3),
497509.
Campbell, J.Y., Lo, A.W., MacKinlay, A.C., 1997. The Econometrics of Financial
Markets. Princeton University Press, Princeton, NJ.
Central Intelligence Agency, 2009. The World Handbook. <https://www.cia.gov/
library/publications/download/> (accessed 06.04.09).
Chamberlain, G., 1978. Omitted variable bias in panel data: estimating the returns
to schooling. Annales de LInsee 3031, 4982.
Crack, T.F., 1999. A classic case of data snooping for classroom discussion. The
Journal of Financial Education 25, 9297, Fall.
Daniel, W.W., 1978. Applied Nonparametric Statistic. Houghton Mifin Co., Boston,
MA.
Daradkeh, T.K., 1992. Parasuicide during Ramadan in Jordan. Acta Psychiatrica
Scandinavica 86 (3), 253254.
Dowling, M., Lucey, B.M., 2005. Weather, biorhythms, beliefs and stock returns
some preliminary Irish evidence. International Review of Financial Analysis 14
(3), 337355.
Duo, E., Saez, E., 2002. Participationininvestment decisions ina retirement plan: the
inuence of colleagues choices. Journal of Public Economics 85 (1), 121148.
Edmans, A., Garcia, D., Norli, O., 2007. Sports sentiment and stock returns. Journal of
Finance 62 (4), 19671998.
Edwards, R.D., 2008. Health risk and portfolio choice. Journal of Business &
Economic Statistics 26 (4), 472485.
Ellision, C.G., Burdette, A.M., Hill, T.D., 2009. Blessed assurance: religion, anxiety,
and tranquility among US adults. Social Science Research 38 (3), 656667.
Fama, E.F., 1998. Market efciency, long-term returns, and behavioral nance.
Journal of Financial Economics 49 (3), 283306.
Frieder, L., Subrahmanyam, A., 2004. Non-secular regularities in returns and
volume. Financial Analysts Journal 60 (4), 2934.
Fuhrman, J., 1998. Fasting and Eating for Health. A Medical Doctors Program for
Conquering Disease. St. Martins Press, New York, NY.
Gibbons, M.R., Hess, P., 1981. Day of the week effects and asset returns. Journal of
Business 54 (4), 579596.
Gultekin, M.N., Gultekin, N.B., 1983. Stock market seasonality. Journal of Financial
Economics 12 (4), 469481.
Haugen, R.A., Baker, N.L., 1996. Commonality in the determinants of expected stock
returns. Journal of Financial Economics 41 (3), 401439.
Hilary, G., Hui, K.W., 2009. Does religion matter in corporate decision making in
America? Journal of Financial Economics 93 (3), 455473.
Hirshleifer, D., 2001. Investor psychology and asset pricing. Journal of Finance 56
(4), 15331598.
Hirshleifer, D., Shumway, T., 2003. Good day sunshine: stock returns and the
weather. Journal of Finance 58 (3), 10091032.
Hong, H., Kacperczyk, M., 2009. The price of sin: the effects of social norms on
markets. Journal of Financial Economics 93 (1), 1536.
Hong, H., Kubik, J.D., Stein, J.C., 2004. Social interaction and stock-market
participation. Journal of Finance 59 (1), 137163.
Jaffe, J.F., 1974. Special information and insider trading. Journal of Business 47 (3),
410428.
Kamstra, M.J., Kramer, L.A., Levi, M.D., 2000. Losing sleep at the market: the daylight
saving anomaly. American Economic Review 90 (4), 10051011.
Kanji, G.K., 1993. 100 Statistical Tests. SAGE Publications, Newbury Park, CA.
Keim, D.B., 1983. Size-related anomalies and stock return seasonality. Journal of
Financial Economics 12 (1), 1332.
Keim, D.B., Stambaugh, R.F., 1984. A further investigation of the weekend effect in
stock prices. Journal of Finance 39 (3), 819835.
Knerr, I., Pearl, P.L., 2008. Ketogenic diet: stoking energy stores and still posing
questions. Experimental Neurology 11, 1113.
Kothari, S.P., Warner, J.B., 2007. Econometrics of event studies. In: Eckbo, Espen, B.
(Eds.), Handbook of Corporate Finance: Empirical Corporate Finance. Elsevier/
North-Holland, Amsterdam.
Lakonishok, J., Smidt, S., 1988. Are seasonal anomalies real? A ninety-year
perspective. Review of Financial Studies 1 (4), 403425.
Liu, W., 2006. A liquidity-augmented capital asset pricing model. Journal of
Financial Economics 82 (3), 631671.
Loewenstein, G., Weber, E.U., Hsee, C.K., Welch, N., 2001. Risk as feelings.
Psychological Bulletin 127 (2), 267286.
Lucey, B., Dowling, M., 2005. The role of feelings in investor decision-making.
Journal of Economic Surveys 19 (2), 211239.
Mitchell, M.L., Stafford, E., 2000. Managerial decisions and long-term stock price
performance. Journal of Business 73 (3), 287329.
Perk, G., Ghanem, J., Aamar, S., Ben-Ishay, D., Bursztyn, M., 2001. The effect of the
fast of Ramadan on ambulatory blood pressure in treated hypertensives. Journal
of Human Hypertension 15, 723725.
Rosen, H.S., Wu, S., 2004. Portfolio choice and health status. Journal of Financial
Economics 72 (3), 457484.
Rozeff, M.S., Kinney Jr., W.R., 1976. Capital market seasonality: the case of stock
returns. Journal of Financial Economics 3 (4), 379402.
Saleh, S.A., Elsharouni, S.A., Cherian, B., Mourou, M., 2005. Effects of Ramadan
fasting on waist circumference, blood pressure, lipid prole, and blood sugar on
a sample of healthy Kuwaiti men and women. Malaysian Journal of Nutrition 11
(2), 143150.
Seyyed, F., Abraham, A., Al-Hajji, M., 2005. Seasonality in stock returns and
volatility: the Ramadan effect. Research in International Business and Finance
19 (3), 374383.
Shiller, R.J., 1981. Do stock prices move too much to be justied by subsequent
changes in dividends? American Economic Review 76, 483498.
Sias, R.W., Starks, L.T., 1997. Institutions and individuals at the turn-of-the-year.
Journal of Finance 52 (4), 15431562.
Simon, H.A., 1967. Motivational and emotional controls of cognition. Psychological
Review 74 (1), 2939.
Stulz, R.M., Williamson, R., 2003. Culture, openness and nance. Journal of Financial
Economics 70 (3), 313349.
Subrahmanyam, A., 2007. Behavioural nance: a review and synthesis. European
Financial Management 14 (1), 1229.
Wagster, J.D., 1996. Impact of the 1988 Basle accord on international banks. Journal
of Finance 51 (4), 13211346.
Weber, M., 1905. The Protestant Ethic and the Spirit of Capitalism. Allen & Unwin,
London.
White, H., 1980. A heteroscedasticity-consistent covariance matrix estimator and a
direct test for heteroscedasticity. Econometrica 48 (4), 817838.
Wright, W.F., Bower, G.H., 1992. Mood effects on subjective probability assessment.
Organizational Behavior and Human Decision Processes 52, 276291.
Yamamoto, H., 1980. An electroencephalographical study of fasting therapy with
special reference to EEG power spectral changes. Japanese Journal of
Psychosomatic Medicine 20 (4), 325335.
Yamamoto, H., Suzuki, J., Yamauchi, Y., 1979. Psychophysiological study on fasting
therapy. Symposiumon Strategies inPsychosomatic Practice and Research, Bodo,
Norway, 1978, Psychotherapy and Psychosomatics vol. 32 (14), 229240.
Zellner, A., 1962. An efcient method of estimating seemingly unrelated regression
equations and tests for aggregation bias. Journal of the American Statistical
Association 57, 348368.
J. Biakowski et al. / Journal of Banking & Finance 36 (2012) 835845 845